Home › GPUs & Macs › NVIDIA RTX 5080 vs NVIDIA RTX 4090
NVIDIA RTX 5080 vs NVIDIA RTX 4090 for local LLMs
Across 5 tests measured on both, the NVIDIA RTX 4090 is faster in 4 and generates a median 1.68× the tokens/s of the NVIDIA RTX 5080.
On Llama 2 7B (Q4_0, llama-bench) the NVIDIA RTX 5080 generates 192 tokens/s and the NVIDIA RTX 4090 190 tokens/s (measured).
Same model, same quant, same test
| Model | Quant / test | RTX 5080 tok/s | RTX 4090 tok/s | Faster | Sources and run details |
|---|---|---|---|---|---|
| Llama 2 7B | Q4_0 llama-bench | 192 Vulkan | 190 Vulkan | even | llama.cpp #10879 RTX 5080 · llama.cpp #10879 RTX 4090 |
| gpt-oss 20B | MXFP4 llama-bench | 229 CUDA | 255 CUDA | RTX 4090 +11% | llama.cpp #15396 RTX 5080 · llama.cpp #15396 RTX 4090 |
| Llama 3.1 8B Instruct | Q4_K_M LocalScore | 35.5 llamafile | 105 llamafile | RTX 4090 +197% | LocalScore RTX 5080 · LocalScore RTX 4090 |
| Llama 3.2 1B Instruct | Q4_K_M LocalScore | 140 llamafile | 292 llamafile | RTX 4090 +109% | LocalScore RTX 5080 · LocalScore RTX 4090 |
| Qwen2.5 14B Instruct | Q4_K_M LocalScore | 29.3 llamafile | 49.1 llamafile | RTX 4090 +68% | LocalScore RTX 5080 · LocalScore RTX 4090 |
Generation speed in tokens/s, best measured backend for each chip (shown under the number). Backends can differ between the two chips; check them before drawing conclusions. Measured results only, nothing estimated.
Specs
| NVIDIA RTX 5080 | NVIDIA RTX 4090 | |
|---|---|---|
| Memory | 16 GB | 24 GB |
| Memory bandwidth | 960 GB/s | 1008 GB/s |
All results for the NVIDIA RTX 5080 · All results for the NVIDIA RTX 4090 · More comparisons